pattern in history

Le 11.04.2006 14:12, Prof Brian Ripley a ?crit :

That's an interesting suggestion, but

1) it is not a good idea to change the order of arguments in a function.

Yes, my mystake. It's because i like to call 'history(something)' 
directly and not 'history(pattern=something)'.
I hadn't in mind the possibility to break existing code.

2) is the 'unique' argument useful?  I cannot see it being used if
there is no pattern search, nor I do see the merit in showing repeated 
lines if I have subselected.

3) like ls(), testing if 'pattern' were missing or NULL would be a 
better idea.

Right.
There is also a problem with the order of the calls.
If I do :

 > ls()
 > l <- mean(rnorm(50))
 > ls()
 > history(pattern="^l")
it prints :
ls()
l <- mean(rnorm(50))

when it should print :
l <- mean(rnorm(50))
ls()

We can use rev twice (like in history3 below), but is it worth it ?

history3 <-
function (max.show = 25, reverse = FALSE, pattern, ...)
{
   file1 <- tempfile("Rrawhist")
   savehistory(file1)
   rawhist <- scan(file1, what = "", quiet = TRUE, sep = "\n")
   if(!missing(pattern)) rawhist <- rev( unique( 
rev(rawhist[grep(pattern, rawhist, ...)] ) ) )
   unlink(file1)
   nlines <- length(rawhist)
     inds <- max(1, nlines - max.show):nlines
   if (reverse)
       inds <- rev(inds)
   file2 <- tempfile("hist")
   write(rawhist[inds], file2)
   file.show(file2, title = "R History", delete.file = TRUE)
}

visit the R Graph Gallery : http://addictedtor.free.fr/graphiques
mixmod 1.7 is released : http://www-math.univ-fcomte.fr/mixmod/index.php
+---------------------------------------------------------------+
| Romain FRANCOIS - http://francoisromain.free.fr               |
| Doctorant INRIA Futurs / EDF                                  |
+---------------------------------------------------------------+